Optimizing a Military Supply Chain in the Presence of Random, Non-Stationary Demands

Demand for supplies, such as ammunition, during a military operation is a scenario-dependent random variable that may be subject to high variance. The challenge is to design an efficient military logistics supply chain that satisfies uncertain, non-stationary demands, while taking into account the volatility and singularity of military operations. This research focuses on the development of a modeling framework that determines the optimal deployment of transportation assets and supplies at the operational level, with possible interdiction by enemy forces. We term this model, Optimal Military Logistics Supply Chain (OPTiMiLSC). This is a two-level, multiple time period scenario-based stochastic model. OPTiMiLSC uses a combination of optimization, scenario-based simulation and statistical analysis. We use a “scenario tree” method to generate the demand scenarios. The results show a positive correlation between the number of demand scenarios and the probability that a random demand scenario is satisfied. We compare OPTiMiLSC with two deterministic optimization approaches. The first approach is where demands are fixed at the 90th percentile, which tends to over-supply when compared to OPTiMiLSC. The mean value approach, on the other hand, tends to under-supply. OPTiMiLSC enables military planners to establish a robust logistic plan that responds more adequately to an intra-theater operation.